From 63eb99e0d348e28df20cce1f64607c4ea049e372 Mon Sep 17 00:00:00 2001
From: Zef Hemel <zef@zef.me>
Date: Sat, 2 Mar 2024 13:54:31 +0100
Subject: [PATCH] Fixes #771

---
 plugs/index/header.ts   | 72 ++++++++++++++++++++++++++++-------------
 website/Space Script.md |  2 +-
 2 files changed, 51 insertions(+), 23 deletions(-)

diff --git a/plugs/index/header.ts b/plugs/index/header.ts
index fb080f5c..61fe46a7 100644
--- a/plugs/index/header.ts
+++ b/plugs/index/header.ts
@@ -1,34 +1,62 @@
-import { collectNodesMatching } from "$sb/lib/tree.ts";
+import {
+  collectNodesMatching,
+  collectNodesOfType,
+  renderToText,
+} from "$sb/lib/tree.ts";
 import type { CompleteEvent, IndexTreeEvent } from "../../plug-api/types.ts";
 import { ObjectValue } from "../../plug-api/types.ts";
 import { indexObjects, queryObjects } from "./api.ts";
 import { parsePageRef } from "$sb/lib/page_ref.ts";
+import { extractAttributes } from "$sb/lib/attribute.ts";
 
-type HeaderObject = ObjectValue<{
-  name: string;
-  page: string;
-  level: number;
-  pos: number;
-}>;
+type HeaderObject = ObjectValue<
+  {
+    name: string;
+    page: string;
+    level: number;
+    pos: number;
+  } & Record<string, any>
+>;
 
 export async function indexHeaders({ name: pageName, tree }: IndexTreeEvent) {
   const headers: ObjectValue<HeaderObject>[] = [];
 
-  collectNodesMatching(tree, (t) => !!t.type?.startsWith("ATXHeading")).forEach(
-    (n) => {
-      const level = +n.type!.substring("ATXHeading".length);
-      const name = n.children![1].text!.trim();
-      headers.push({
-        ref: `${pageName}#${name}@${n.from}`,
-        tag: "header",
-        level,
-        name,
-        page: pageName,
-        pos: n.from!,
-      });
-    },
-  );
-  // console.log("Found", headers.length, "headers(s)");
+  for (
+    const n of collectNodesMatching(
+      tree,
+      (t) => !!t.type?.startsWith("ATXHeading"),
+    )
+  ) {
+    const level = +n.type!.substring("ATXHeading".length);
+    const tags = new Set<string>();
+
+    collectNodesOfType(n, "Hashtag").forEach((h) => {
+      // Push tag to the list, removing the initial #
+      tags.add(h.children![0].text!.substring(1));
+      h.children = [];
+    });
+
+    // Extract attributes and remove from tree
+    const extractedAttributes = await extractAttributes(
+      ["header", ...tags],
+      n,
+      true,
+    );
+    const name = n.children!.slice(1).map(renderToText).join("").trim();
+
+    headers.push({
+      ref: `${pageName}#${name}@${n.from}`,
+      tag: "header",
+      tags: [...tags],
+      level,
+      name,
+      page: pageName,
+      pos: n.from!,
+      ...extractedAttributes,
+    });
+  }
+
+  console.log("Found", headers, "headers(s)");
   await indexObjects(pageName, headers);
 }
 
diff --git a/website/Space Script.md b/website/Space Script.md
index 8b957578..1abb0f01 100644
--- a/website/Space Script.md	
+++ b/website/Space Script.md	
@@ -140,7 +140,7 @@ SilverBullet indexes various types of content as [[Objects]]. There are various
 The `silverbullet.registerAttributeExtractor` API takes two arguments:
 
 * `def` with currently just one option:
-  * `tags`: Array of tags this extractor should be applied to, could be a built-in tag such as `item`, `page` or `task`, but also any custom tags you define.
+  * `tags`: Array of tags this extractor should be applied to, could be a built-in tag such as `item`, `page`, `paragraph`, `header`, or `task`, but also any custom tags you define.
 * `callback`: the callback function to invoke (can be `async` or not). This callback is passed the following arguments:
   * `text`: the text of the object to extract attributes for
   * return value: an object of attribute mappings, possibly overriding built-in ones.